Deep earth structure

Abstract

Major developments in earth structure in the last four years have been concentrated in the description of the earth's lateral heterogeneity: the regions that are heterogenous and the per cent variation of velocity and density in each region. Most studies find that lateral variation is concentrated in the upper 400 and lower 200 km. of the mantle. A radially symmetric earth model has been defined that represents the best average fit to seismic data in a broad frequency band, sampling many regions. P and S velocity is found to increase in zones of 50 km. or less at 400 and 650 to 700 km. depth. The model is transversely anisotropic in the upper mantle. It possesses a vertical axis of symmetry such that the elastic constants are different for vertical propagation than for horizontal and intermediate angles of propagation. The real earth generally exhibits azimuthal anisotropy as well, but the azimuthal anisotropy cannot be resolved by a global average of data. The nature and magnitude of the anisotropy agrees with that found in ultramafic samples of the upper mantle. In attenuation, models of intrinsic attenuation have included the dispersive properties of intrinsic anelasticity and constructed relaxation models consistent with an observed frequency dependence of Q in the body wave band. There has been progress in mapping the scattering properties of the lithosphere. Attenuation due to scattering in the crust and lithosphere has been shown to have strong effects on the amplitudes of seismic waves at local and teleseismic distances.

Decay of wave fields near an explosive source due to high‐strain, nonlinear attenuation

“…We may generalize the rheological model and yet keep it linear by assuming that Q varies with frequency in the band of interest. This assumption is supported by the results of a variety of observational and theoretical seismological studies recently reviewed by Cormier [1983], by Kirby [1983], and, for the class of problems examined here, by Bache [1985]. Ad-ditional support for such Q models stems from a rapidly growing body of laboratory data [e.g., Winklet and Nut, 1979' Tittman et al, 1981' Bulau et al, 1984' Ulug and Berckherner, 1984.…”